Scientia Iranica A (2012) 19 (6), 1411–1421

Sharif University of Technology

Scientia Iranica
Transactions A: Civil Engineering
www.sciencedirect.com

Bearing failure modes of rock foundations with consideration of

joint spacing
M. Imani a , A. Fahimifar a,∗ , M. Sharifzadeh b
a
Department of Civil and Environmental Engineering, Amirkabir University of Technology, Tehran, Iran
b
Department of Mining and Metallurgy Engineering, Amirkabir University of Technology, Tehran, Iran

Received 18 October 2011; revised 29 June 2012; accepted 7 August 2012

KEYWORDS
Jointed rock foundation;
Bearing capacity;
General shear failure;
Excessive deformation;
Joint spacing.
Abstract In this study, two main failure modes of jointed rock foundations consisting general shear failure
and failure due to excessive deformation are discussed. For each of these two modes, the method of
determining ultimate bearing capacity is presented and the effect of joint spacing is also examined. For
the former mode, an upper bound method of limit analysis is employed, while for the latter one, distinct
element numerical modeling is used to investigate the effect of joint spacing on the bearing capacity. The
calculations are performed for a rock mass containing two orthogonal joint sets that the orientation angle
of the first joint set is 15°, 30° and 45° to the horizontal, respectively. A non-dimensional parameter called
‘‘Spacing Ratio’’ (SR) is used to examine the effect of different joint spacing. This study shows that in the
case of general shear failure mode, joint spacing does not have significant effect on the bearing capacity.
In the failure mode induced by excessive deformation of the rock mass, for SR < 30, increasing the SR
results in decreasing the bearing capacity, however, for SR > 30, the joint spacing does not significantly
affect the bearing capacity.
© 2012 Sharif University of Technology. Production and hosting by Elsevier B.V.
Open access under CC BY license.

1. Introduction those derived through considering the jointed rock mass as a

The bearing capacity of a strip footing resting on soil has
been widely studied in the literature. Because of inhomogeneity
and discontinuity of rock masses, few attempts have been made
to study the bearing capacity of rock foundations, particularly
jointed rocks. Among them, the researches performed by
Serrano and Olalla [1,2], Sutcliffe et al. [3], Yang and Yin [4],
Merifield et al. [5] and Saada et al. [6] are concerned with
rock masses obeying Hoek–Brown failure criterion which is
applicable for intact or heavily crushed rocks. Moreover, using
the upper bound method of limit analysis, Maghous et al. [7]
assessed the load bearing capacity of rock foundations resting
on a regularly jointed rock. They considered the rock matrix
and the joints separately, and compared the obtained results by
∗
Corresponding author. Tel.: +98 2164543011; fax: +98 2164543011.
E-mail address: Fahim@aut.ac.ir (A. Fahimifar).
Peer review under responsibility of Sharif University of Technology.
1026-3098 © 2012 Sharif University of Technology. Production and hosting by Elsevier B.V. Open access under CC BY license.
doi:10.1016/j.scient.2012.10.007
homogenized medium. Sutcliffe et al. [3] analyzed the bearing
capacity of rock masses containing one to three sets of closely
spaced joints. Also, Halakatevakis and Sofianos [8] analyzed a
series of jointed rock samples containing one to three joint
sets with various spacing and dip angles using the distinct
element code, UDEC. They have concluded that the strength
of the models is independent of the joint spacing. No further
published documents regarding the effects of joint spacing on
the bearing capacity of rock foundations have been found.
Depending on the failure mode of rock foundations, spacing
of joints has different influences on the bearing capacity. In
practice, two main failure modes may occur which are named as
general shear failure and punching shear failure. The latter can
be attributed as the failure due to excessive deformation. Most
of the above mentioned methods for determining rock bearing
capacities are based on the general shear failure mode.
In this paper, the two mentioned failure modes are discussed
precisely for a rock mass containing two orthogonal joint sets.
The effect of joint spacing on the bearing capacity of rock
foundations in each of these failure modes is also investigated.
For the general shear mode, an upper bound method of
limit analysis is employed and the ultimate bearing capacity
1412 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421

relatively uninvolved, so the movement on both sides of the

Nomenclature footing will be minimal. In the case of soil beddings, latter
type of failure is typically observed in compressible soils. The
B foundation width failure due to excessive deformation may also take place in
ci cohesion of intact rock rock masses but often not as a result of compressibility of the
cj cohesion of joint set mass, because even weak rock masses are not as compressible
D total energy dissipation as most types of soils. As emphasized by Eurocode 7 [10], one
DEM distinct Element Method
of the different types of limit states for a rock foundation is the
Kn joint normal stiffness
excessive deformation of the ground in which the strength of
Ks joint shear stiffness
rock is significant. In such a case, the mobilized bearing pressure
n joint sets number
continuously increases with the footing movement, apparently
n0 an integer number
without reaching an ultimate bearing capacity, at least within
Ncj , Nci , Nq and Nγ bearing capacity coefficients
the bounds of small displacement approach. Therefore, the slip
q surcharge
surfaces do not reach the ground surface and a large settlement
qu ultimate bearing capacity of dry rock
occurs in the rock mass. In this case, the methods of determining
SR spacing ratio
the bearing capacity based on the general shear failure may
Si spacing of the ith joint set
not accurately result in the critical load. A proper approach
UBS upper bound solution proposed in this study
for determining the critical load in such cases is to use the
W total work of external forces
load–settlement curve of the rock foundation.
α orientation angle of one of the joint sets with
In practical situations, depending on the rock mass proper-
horizontal direction
ties, the failure due to excessive deformation may occur prior to
γ unit weight of rock mass
the general shear failure which reveals the necessity of taking
θ angle between the velocity discontinuity line
care of deformations. In the following sections, the procedure
passing through intact rock with horizontal
used for determining ultimate bearing capacity in each of the
direction
two mentioned failure modes is investigated and the effect of
ϕi friction angle of intact rock
joint spacing is discussed.
ϕj friction angle of joint set

2.1. General shear failure

equation is derived in which the joint spacing can be taken
into account. Orientation angles equal to 15°, 30° and 45° were
selected for one of the joint sets, while the second joint set
was assumed to be perpendicular to the first set. For the failure
mode, due to excessive deformation, Distinct Element Method
(DEM) was used to investigate the effect of joint spacing on
the bearing capacity. In this regard, the concept of ‘‘spacing
ratio’’ (SR) initially proposed by Serrano and Olalla [9] was used
to introduce joint spacing into the discontinuum numerical
models. The non-dimensional parameter, SR, is expressed as
follows:
n
 1
SR = B , (1)
i=1
Si
where, B is the footing width, Si is the spacing of the ith joint set
and n is the number of joint sets.
Using load–settlement curve obtained from the numerical
analyses, the ultimate bearing capacity was obtained for
different SRs; and the particular SR, in which the rate of
variation of the bearing capacity becomes negligible, was
selected as the limit of joint spacing effectiveness. This limiting
value is called as the critical spacing ratio (SRcr ). Finally, a
procedure for determining allowable bearing capacity of rock
masses was proposed in this paper which can be used in
practical applications.
2. Modes of bearing failure of rock foundations
In the case of general shear failure, a continuous shear failure
occurs in the rock bedding from beneath the footing to the
ground surface and results in pushing up the bedding on both
sides of the footing. But in failure, due to excessive deformation
(punching shear failure), the failure takes place beneath the
footing and the rock mass outside the loaded area remains
Upper bound theorem of limit analysis was applied to obtain
the bearing capacity of a rock mass subjected to the general
shear failure. According to the theorem, the rate of energy
dissipation is not less than the rate of work of the external forces
for any kinematically admissible failure mechanism.
Construction of an admissible failure mechanism is the first
step in solving the bearing capacity problem using the upper
bound method. In the foundations with centric and vertical
loadings, the failure mechanism is usually considered to be
symmetrical (i.e. a two-sided mechanism). However, in jointed
rock masses, the failure mechanism may be affected by the
joint sets, and thus converted to an asymmetrical shape. For
determining the shape of the failure mechanism, numerical
analyses were performed using the distinct element code,
UDEC. Since the upper bound formulation obtained here is valid
only for the case of general shear failure, the properties of intact
rock and joint sets were selected such that the failure surfaces in
the rock mass reach the ground surface (general shear failure).
The numerical modeling was performed for a flexible
foundation with one meter width located on the surface of
a jointed rock mass containing two orthogonal joint sets in
which orientation angles equal to α = 15°, 30° and 45°
were considered for one of the joint sets. It was assumed
that the joints are starting from the foundation extremities. In
the analyses, different SR values, at the range of 3–50, were
examined. The rock blocks were assumed to be deformable and
Mohr Coulomb failure criterion was used for the rock material
and the joint sets.
As an example, for the intact rock cohesion (ci ) equal to
5 MPa, joint sets cohesion (cj ) equal to 50 kPa and intact
rock and joint sets friction angles (ϕi and ϕj , respectively)
equal to 35°, the displacement vectors of the rock blocks are
shown in the left hand side of Figure 1. For α = 45°, the
mechanism is symmetrical (a two-sided mechanism), while for
 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421
Figure 1: Displacement vectors obtained using distinct element method for the case of SR = 10 and the overall shape of the failure mechanisms.
α = 30° and 15°, the mechanism is asymmetrical (a one-
sided mechanism). Moreover, keeping the above-mentioned
properties unchanged, the effect of joint spacing on the
configuration of failure mechanism was also investigated and
no remarkable effect of joint spacing on the shape of the failure
mechanism was observed. Hence, a two-sided symmetrical
failure mechanism (TS failure mechanism) was used for the case
of α = 45° and a one-sided asymmetrical failure mechanism
(OS failure mechanism) was used for α = 15° and 30°. For
obtaining the least magnitude of upper bound bearing capacity,
two different TS mechanisms (named TS1 and TS2) and also
two different OS mechanisms (named OS1 and OS2) were
considered as shown in the right hand side of Figure 1.
2.1.1. Failure mechanisms TS1 and OS1
The mechanisms TS1 and OS1 and the corresponding
hodographs are shown in Figure 2. In order to minimize
the maximum internal energy dissipation, and thus to obtain
minimum values for the upper bound bearing capacity, the
greatest possible length of velocity discontinuity lines have
been passed along the joints. It is assumed that both the
beginning and the end points of line CD are located at the
junction of the two joint sets, not within the intact rock block.
For satisfaction of this assumption, CD was considered as a
straight line. In the case of assuming a logarithmic-spiral curve
for CD (such as the one considered in Prandtl mechanism), it is
possible that point D does not locate at the junction of the two
joint sets.
For TS1, since the movements are symmetrical about the
footing vertical axis, it is enough to consider the movement
through the half of the mechanism.
Rock masses may not obey the associated flow rule.
However, in a large number of stability problems such as
1413
bearing capacity, deformation conditions are not so restrictive
and the rock deformation properties do not greatly affect the
collapse load. Therefore, the adoption of the associated flow rule
appears to be reasonable in the limit analysis [11]. According to
the normality rule, the velocity on every discontinuity line must
be inclined at an angle ϕ with that line, where ϕ is the friction
angle of the medium in which the discontinuity line lays (either
ϕi or ϕj ). All the velocities of the mechanism determined in this
way, constitute a kinematically admissible velocity field.
In Figure 2, the foundation width (B), the spacing of the joints
(S1 ), α , ϕi and ϕj are known. Since the absolute magnitude of
the velocity has no influence on the final results, it is assumed
that the magnitude of V0 is equal to unity, i.e. V0 = 1. From
the geometrical relations in the hodographs shown in Figure 2,
the magnitudes of other velocity vectors can be calculated as
follows:
For the mechanism TS1, see Box I.
For the mechanism OS1, see Box II.
The angle of line CD with the horizontal direction (θ ) is:
 
n0 S1
θ = tan−1 − α, (10)
B cos α
where n0 is an integer number which by being multiplied by S1 ,
the length of velocity discontinuity line BD is obtained. So, the
only unknown parameter of the failure mechanisms is n0 .
The rate of energy dissipation (DL ) along each velocity
discontinuity line is:
DL = c · V · cos φ, (11)
where c is the cohesion (either ci or cj ), and V is the velocity
magnitude that makes the angle ϕi or ϕj with a velocity
1414 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421

Figure 2: Failure mechanisms and the corresponding hodographs (a) TS1 and (b) OS1.

cos2 (φi − θ )
V01 = V0 · , (2)
sin α − φj − sin (φi − θ) · cos α + θ − φi − φj
  

− sin α − φj · sin (φi − θ ) + cos α + θ − φi − φj

   
V1 = V0 ·  , (3)
sin α − φj − sin (φi − θ ) · cos α + θ − φi − φj
  

sin2 α + θ − φi + φj
 
V12 = V1 · , (4)
sin 2φj · cos α + θ − φi + φj + sin α + θ − φi − φj
    

sin 2φj + cos α + θ − φi + φj · sin α + θ − φi − φj

     
V2 = V1 · . (5)
sin 2φj · cos α + θ − φi + φj + sin α + θ − φi − φj
    

Box I

cos2 α + θ − φi + φj
 
V01 = V0 ·  , (6)
cos α + θ − φi − φj · sin α + θ − φi + φj − sin 2φj
   

cos α + θ − φi − φj − sin α + θ − φi + φj · sin 2φj

     
V1 = V0 ·  , (7)
cos α + θ − φi − φj · sin α + θ − φi + φj − sin 2φj
   

sin2 α + θ − φi + φj
 
V12 = V1 · , (8)
sin 2φj · cos α + θ − φi + φj + sin α + θ − φi − φj
    

sin 2φj + cos α + θ − φi + φj · sin α + θ − φi − φj

     
V2 = V1 · . (9)
sin 2φj · cos α + θ − φi + φj + sin α + θ − φi − φj
    

Box II

discontinuity line. Thus, the total energy dissipation (D) in the The total external work (W ) in the mechanism TS1 is:
mechanism TS1 is:
W = WABC + 2 × WBCD + WBDE + Wq + Wqu .
 
(14)
D = 2 × (DBC + DCD + DBD + DDE ) . (12)
In the mechanism OS1 it changes to: And, in the mechanism OS1 it changes to:

D = DAC + DBC + DCD + DBD + DDE , (13) W = WABC + WBCD + WBDE + Wq + Wqu , (15)

where DXY is the energy dissipation along the discontinuity line, where WXYZ is the external work of wedge XYZ, Wq is the
XY. external work of surcharge q and Wqu is the external work of
 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421 1415

foundation load which are as follows for TS1 and OS1:

Wqu ,(TS1) = qu · B · V0 , (16)

= qu · B · V0 · cos α + φj .
 
Wqu ,(OS1) (17)
For TS1, by equating Eqs. (12) and (14) and using Eq. (16), the
ultimate bearing capacity for TS1 was obtained as in Box III.
For OS1, by equating Eqs. (13) and (15) and using Eq. (17),
the ultimate bearing capacity for OS1 was obtained as in Box IV. Figure 3: Strip footing on rock masses with two orthogonal joint sets, α = 45°,
Finally, after some rearrangements, the general equation for SR = 3, 6, 15 and 30.
the upper bound of ultimate bearing capacity (qu ) of a shallow
foundation on the assumed jointed rock is obtained as: where:
1 g1 = cos ξ1 sin ξ2 − sin 2φj , (39)
qu = cj Ncj + ci Nci + qNq + γ BNγ , (20)
2 g2 = cos ξ1 − sin ξ2 sin 2φj , (40)
where γ is the total unit weight of the rock mass, Ncj , Nci , Nq and
Nγ are the bearing capacity coefficients that are as follows for g3 = sin 2φj cos ξ2 + sin ξ1 , (41)
TS1 and OS1. Assuming: g4 = sin 2φj + cos ξ2 sin ξ1 . (42)
ξ1 = α + θ − φi − φj , (21)
2.1.2. Failure mechanisms TS2 and OS2
ξ2 = α + θ − φi + φj , (22)
It was not clear prior to the calculations, which mechanism
ξ3 = α + φj , (23) would predict the minimum bearing capacity (optimum upper
ξ4 = α + θ , (24) bound) for the jointed rock foundation. So, in addition to TS1
and OS1, the mechanisms TS2 and OS2 were also considered
ξ5 = φi − θ , (25)
in order not to predefine the failure mechanism. For all
ξ6 = α − φj . (26) cases analyzed, failure mechanisms TS1 and OS1 ensured the
For the mechanism TS1: minimum limit load. For example, for the case of α = 45°, ci =
5 MPa cj = 50 kPa, ϕi = ϕj = 35°, q = 20 kPa, γ = 27 kN/m3
2 cos φj cos α
Ncj = and S1 = S2 = 3.9 cm, the mechanism TS1 yields n0 = 100 and
 f1   qu = 210.8 MPa, while the mechanism TS2 yields n0 = 101 and
f2 f4 qu = 572.3 MPa. Therefore, the mechanisms TS2 and OS2 are
× cos ξ5 + 2
· tan ξ4 sin ξ2 + 2
, (27)
f3 tan α not presented here.
f2 2 cos φi cos α
Nci = · , (28) 2.2. Failure due to excessive deformation
f1 cos ξ4
f2 f4 2 sin ξ3 tan ξ4 Most of the existing analytical methods for determining
Nq = · , (29) ultimate bearing capacity of rock foundations are based on the
f1 f3 tan α
 general shear failure of the rock mass. Therefore, they do not
2f2 seem to be proper for the case of a failure induced by excessive
Nγ = cos α · cos α tan ξ4
f1 deformation of the mass. In such cases, load–settlement curve
sin ξ3 tan ξ4
  
f4 is a popular tool in estimating critical load. This curve can be
× sin ξ5 + · − sin α , (30) obtained from loading tests in the field or using numerical
f3 tan α
methods, where the latter was used in this study by applying
where: DEM. Jointed rock foundations containing two orthogonal joint
f1 = sin ξ6 − sin ξ5 cos ξ1 , (31) sets were considered for dealing with the problem of failure
f2 = − sin ξ6 sin ξ5 + cos ξ1 , (32) due to excessive deformation. Figure 3 shows the general
configuration of some of the constructed models for α = 45°.
f3 = sin 2φj cos ξ2 + sin ξ1 , (33)
In the case of failure due to excessive deformation, no
f4 = sin 2φj + cos ξ2 sin ξ1 . (34) marked change in settlement is taken place in the load–
For the mechanism OS1: settlement curve. In such cases, the stress corresponding to
a settlement equal to 10% of the foundation width is often
cos φj cos α cos2 ξ2

defined as the bearing capacity (the 0.1B method). Maghous
Ncj = tan α +
cos ξ3 g1 et al. [12] and Imani et al. [13] showed that this method may
appropriately estimate the critical load of a rock foundation.

g2  g4 
+ · tan ξ4 sin ξ2 + 2
, (35) Hence, this method was employed in the present study for
g1 g3 tan α
determining the bearing capacity in the case of failure due to
g2 cos φi cos α excessive deformation.
Nci = · , (36)
g1 cos ξ3 cos ξ4
g2 g4 tan ξ3 tan ξ4 3. Results
Nq = · , (37)
g1 g3 tan α
cos α tan ξ4

g2 3.1. General shear failure
Nγ = cos α ·
g1 cos ξ3
g4 sin ξ3 tan ξ4
  
The optimum upper bound bearing capacity is obtained
× sin ξ5 + · − sin α , (38)
g3 tan α by minimizing Eq. (20) with respect to the only unknown
1416 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421

2 × (DBC + DCD + DBD + DDE ) − WABC − 2 × WBCD + WBDE + Wq

 
qu,(TS1) = (18)
B · V0

Box III

 
DAC + DBC + DCD + DBD + DDE − WABC + WBCD + WBDE + Wq
qu,(OS1) = (19)
B · V0 · cos α + φj
 

Box IV

parameter, n0 . Two computer programs were prepared in

MATLAB code to solve the bearing capacity equations for TS1
and OS1. The genetic algorithm provided in the code was used
for minimization. The minimization procedure was performed
subject to the following constraints for OS1:
π π
α+θ < , 0 < α + θ − φi − φj < , n0 > 0. (43)
2 2
And for TS1, the constraint 0 < α − θ + φi + φj < π/2 was
applied in addition to the above constraints.

3.1.1. Comparison with other solutions

Figure 4 represents a comparison among the results of the
proposed Upper Bound Solution (UBS) for the case of SR = 50,
the numerical lower bound solution by Sutcliffe et al. [3] and the
displacement finite element solution by Alehossein et al. [14].
The results are presented as qu /ci ratio versus α for the case
of a footing resting on the surface of a weightless rock mass
Figure 4: Comparison of qu /ci versus α obtained using Sutcliffe et al. [3] and
in which, cj /ci = 0.1 and ϕi = ϕj = 35°. As expected, the
Alehossein et al. [14] and the UBS.
upper bound results are higher than the other two methods,
especially higher than the results of the lower bound solution.
For ponderable rock masses, Figures 7 and 8 show the qu /ci
As discussed previously, in the mechanisms TS1 and OS1, most
ratio versus SR for a ponderable rock mass with γ = 27 kN/m3
of the velocity discontinuity lines passed through joint surfaces,
and other properties depicted in the figures. SR value has also
the position of which was predefined. So, it was not possible
no pronounced effect on the bearing capacity in these cases.
to optimize their location and the minimization procedures of
the mechanisms were performed only for one discontinuity
3.1.3. Bearing capacity coefficients
line (CD). As a result, the differences between the results of
the UBS and other methods were become large. Moreover, in Table 1 presents Ncj , Nci , Nq and Nγ coefficients required
the UBS, there is a jump between the qu /ci for α = 30° and for determining the ultimate bearing capacity of jointed rocks
α = 45°, which is due to the conversion from the one-sided using the UBS (see Eq. (20)). According to the previous section,
since the SR ratio does not have significant effect on the bearing
failure mechanism in α = 30° to the two-sided in α =
capacity, it was assumed that SR = 50 in providing Table 1. This
45°. Moreover, in the proposed approach, when the value of
table can be used to determine the ultimate bearing capacity
‘‘α + θ − ϕi − ϕj ’’ (the angle between the velocity vectors
coefficients for practical cases.
V1 and V01 shown in Figure 2(a) and (b) becomes small, these
velocity vectors become large, resulting in an increase in the
bearing capacity. This phenomenon occurred for the properties 3.2. Failure due to excessive deformation
assumed to obtain the bearing capacity in the case of α = 45°.
Since in the case of failure due to excessive deformation,
the shear surfaces in the material beneath the foundation do
3.1.2. Effect of joint spacing not reach the ground surface, the properties of the intact rock
Using the UBS, the effect of joint spacing was studied for the and the joint sets were selected such that the failure surfaces
general shear failure. The cases examined here, include a footing in the rock mass do not reach the ground surface. In the
resting on the surface of a rock mass, where both the weightless constructed numerical models, it was assumed that ϕi = ϕj =
and ponderable rocks were considered. It should be noted that 35°, joint normal stiffness and joint shear stiffness (Kn and Ks ,
rock masses with different properties were analyzed by DEM respectively) are equal to 100 GPa/m, ci = 20 MPa and cj =
and the displacement vectors obtained showed that the general 2 MPa. Other specifications of the models were similar to those
shear failure occurs for all of the cases. The UBS was applied to described previously for the selection of failure mechanisms for
investigate the effect of joint spacing in this regard. the case of general shear failure.
For weightless rock masses, the qu /ci ratios versus SR are As an example, for the case of α = 45°, Figures 9 and 10
shown in Figures 5 and 6 for the rock mass properties depicted show the settlement contours and the corresponding displace-
in the figures. As observed, SR value does not significantly affect ment vectors at a settlement equal to 10% of the footing width,
bearing capacity. for the case of SR values equal to 6 and 40, respectively. The
 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421 1417

Figure 5: Variation of the qu /ci versus SR for the case of a weightless rock mass with cj /ci = 0 and (a) ϕi = 35°, ϕj = 25° and (b) ϕi = 45°, ϕj = 35°.

Figure 6: Variation of the qu /ci versus SR for the case of a weightless rock mass with cj /ci = 0.1 and (a) ϕi = 35°, ϕj = 25° and (b) ϕi = 45°, ϕj = 35°.

Figure 7: Variation of the qu /ci versus SR for the case of a ponderable rock mass with cj = 0 and ci = 1 MPa and (a) ϕi = 35°, ϕj = 25° and (b) ϕi = 45°, ϕj = 35°.

displacement vectors are predominantly in the downward di-
rection. Although few vectors are at an angle to the vertical,
they do not show a flow pattern beneath the footing up to the
ground surface. Therefore, it can be concluded that for the as-
sumed properties of the intact rock and the joint sets, excessive
deformation (settlement equal to 10% of the footing width) oc-
curs prior to the general shear failure.
Figure 11 shows the variations of the qu /ci ratio obtained
from the 0.1B method, versus SR for different values of α . Ac-
cording to the figure, for SR < 30, increasing the SR results in
decreasing the bearing capacity, but for SR > 30, the joint spac-
ing does not affect the bearing capacity significantly. Hence,
SR = 30 can be taken into account as the approximate limit for
the influence of jointing on the bearing capacity and it is named
in this study as the critical spacing ratio (SRcr ).
DEM results are highly dependent on the selected rock
parameters. For considering the effect of the intact rock and
the joint set properties on the SRcr , sensitivity analyses were
performed. Because of the large number of models, the results of
the case of α = 30° are only presented here. It should be noted
that for all of the assumed properties for the intact rock and
the joint sets in the following sections, excessive deformation
1418 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421

Figure 8: Variation of the qu /ci versus SR for the case of a ponderable rock mass with cj = 0.1 MPa, ci = 1 MPa and (a) ϕi = 35°, ϕj = 25° and (b) ϕi = 45°, ϕj = 35°.

Table 1: Bearing capacity coefficients for jointed rock foundations.

α° ϕi ° ϕj ° Ncj Nci Nq Nγ

20 20 25.65 4.78 12.08 19.89

20 41.17 6.75 19.88 43.26
30
30 81.07 9.02 53.01 155.59
20 74.39 10.47 36.86 110.23 Figure 10: Rock mass with α = 45° and SR = 40. (a) Settlement contour and
15
40 30 186.37 17.88 123.61 547.04 (b) displacement vectors.
35 351.63 24.10 267.44 1558.90
20 168.74 19.69 85.88 387.66
50 30 758.07 60.58 510.87 4452.54
35 3237.01 185.2 2488.29 42,425.89
20 20 17.95 5.45 9.52 12.92
20 28.08 7.68 15.65 28.42
30
30 60.22 11.35 42.32 102.62
30 20 49.45 11.91 28.99 72.63
40 30 136.34 22.50 98.59 360.13
35 275.92 32.60 221.56 1057.94
20 109.39 22.37 67.48 253.92
50 30 547.11 76.27 407.77 2908.24
35 2523.5 251.26 2067.41 28,590.03
20 20 17.62 6.75 9.87 9.75
20 26.73 9.51 16.22 21.44
30
30 71.1 17.78 52.32 89.65
45 20 45.72 14.73 30 54.26
40 30 157.74 35.16 121.57 311.69
35 399.63 64.06 334.57 1106.81
20 98.23 27.59 69.64 187.09
Figure 11: Variation of qu /ci with SR for two joint sets.
50 30 621.50 118.94 501.57 2477.09
35 3620.92 494.36 3125.55 29,415.80

Figure 9: Rock mass with α = 45° and SR = 6. (a) Settlement contour, and
(b) displacement vectors.
occurs in the rock mass and the displacement vectors of the rock
blocks are similar to those presented in Figs. 9(b) and 10(b).
3.2.1. Effect of shear strength properties
Keeping ci unchanged as it was assumed previously, Fig-
ure 12(a) shows the qu /ci versus SR for the case of cj /ci = 0.1,
0.3 and 0.5, while other properties of the rock mass are similar
to those previously considered. Also, Figure 12(b) presents the
qu /ci versus SR for ϕi = ϕj = 25°, 35° and 45° and other prop-
erties as they were assumed previously. The figures reveal that
the SRcr = 30 proposed in this research, does not change with
changes in the cj /ci ratio if the deformation mode at critical load
of the foundation remains unchanged.
3.2.2. Effect of joint stiffness
To examine the influence of joint stiffness, analyses were
performed using new joint normal stiffness and joint shear
stiffness (Knew ) equal to 25% and 50% smaller and larger than the
initial values (Kini = 100 GP/m) and the rest of the parameters
were kept unchanged. According to Figure 13, it is clear that the
stiffness of the joints does not change the SRcr .
 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421 1419

Figure 12: Variation of qu /ci versus SR for (a) cj /ci = 0.1, 0.3 and 0.5 and (b) ϕi = ϕj = 25°, 35° and 45°.

Figure 13: Effect of joint stiffness on the qu /ci versus SR.

Figure 14: Effect of settlement limit on the qu /ci versus SR.
3.2.3. Effect of settlement magnitude at critical load
In this study, it is assumed that the excessive settlement
that occurs at critical load of the rock mass is equal to 10% of
the foundation width (0.1B). For investigating the effect of this
hypothesis on the SRcr , settlements equal to 0.08B and 0.13B
were also examined and the rest of the parameters remained
unchanged as they were assumed previously. Figure 14 reveals
that the settlement limit does not have significant effect on the
SRcr .

3.2.4. Effect of joint set numbers

Analyses were performed for the rock foundations contain-
ing one and three joint sets. Additional models were also ana-
lyzed to investigate the influence of the number of the joint sets
on the SRcr , the configuration of some of which is shown in Fig-
ure 15. All of the intact rock and the joint sets properties were
considered as previously. Figure 16(a) and (b) reveal that the
SRcr for both cases of one and three joint sets is similar to the
condition of a rock foundation with two joint sets.
4. Discussions
Results of this study indicate that for the general shear
failure, increase in ci results in improving the contribution
of intact rock cohesion in bearing capacity which tends to
Figure 15: Strip footing on rock masses with one and three joint sets.
increase the qu /ci ratio. Increase in ϕi and ϕj leads to the
greatest increment in bearing capacity in the case of α =
45°, while the smallest for α = 15°. This is because of
the fact that by increasing the friction angles, the volume of
failure mechanism for α = 45° becomes larger than the
other orientations. The same results were also obtained for
ponderable rocks. In this case, the contribution of the rock
weight results in increasing the bearing capacity. Moreover, for
1420 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421
Figure 16: Variation of qu /ci versus SR for (a) one joint set and (b) three joint sets.
both weightless and ponderable rock masses, the SR ratio (and
hence, the joint spacing) does not have remarkable effect on the
bearing capacity. This conclusion is only valid for the cases in
which the variations of joint spacing do not change the overall
configuration of failure mechanism. In the UBS, as variation of
SR imposes small changes in the location of discontinuity line
CD, it does not change the overall configuration of the failure
mechanisms. As discussed previously, for the rock models
containing one through three joint sets, similar results were
obtained by Halakatevakis and Sofianos [8].
The results obtained for the case of failure due to excessive
deformation show that for SR < 30, increasing SR causes the
qu /ci to decrease significantly. This is because of the fact that
larger number of joints will provide smaller shear resistance
because of the reduced roughness of the rock mass. For SR >
30, the rate of reduction in qu /ci , decreases remarkably, which
implies that the reduction of joint spacing does not affect the
bearing capacity any more. Also, the difference in the qu /ci
ratios for various orientation angles decreases for the case of
SR > 30 which means the small effect of joint spacing on
the bearing capacity for SR > 30. The sensitivity analyses
performed in this study shows that the mechanical properties
of the rock mass do not significantly affect the influence of SR
on the bearing capacity for the case of failure due to excessive
deformation.
Hence, one can say that in the case of general shear fail-
ure, the effect of joint spacing on the bearing capacity is
negligible. But in the failure due to excessive deformation, re-
duction in joint spacing results in reduction of bearing capacity
for SR < 30, while it does not have significant effect in the case
of SR > 30.
5. Conclusions
Two main bearing failure modes of jointed rock foundations
were discussed in this study and the effect of joint spacing
were investigated in each case. It was concluded that in
the case of general shear failure, the joints spacing do not
change the bearing capacity of the assumed jointed rock
foundations significantly, while the bearing capacity will be
affected when the failure is due to excessive deformation of
the rock foundation. In the recent case, for SR < 30, increasing
the SR results in decreasing the bearing capacity, but for SR >
30, the joint spacing does not significantly affect the bearing
capacity.
Since the failure, due to excessive deformation, may occur
prior to the general shear failure, determining the ultimate
bearing capacity of rock foundations only by using the equa-
tions based on the general shear failure of rock foundation, may
lead to unrealistic results. Therefore, in practical applications,
allowable bearing capacity could be determined using the fol-
lowing procedure:
1. Determination of the ultimate bearing capacity using the
methods based on general shear failure of the rock mass (like
the UBS) and applying an appropriate safety factor.
2. Determination of the ultimate bearing capacity using the
methods which are capable of taking into account the failure
induced by excessive deformation (like DEM), and applying
an appropriate safety factor.
3. Determination of the foundation pressure pertinent to the
allowable settlement.
4. Using possibly the minimum value obtained from the three
above stages as the final allowable bearing capacity.
In proceeding to the second stage mentioned above, the
SRcr = 30 proposed in this study is a very useful criterion. It
shows that even for a jointed rock mass with SR > 30, the
ultimate bearing capacity cannot be further decreased for the
SR ratios greater than 30.
For dealing with the effect of joint spacing on the bearing
capacity, specific properties were assumed for the rock mass
in order to form the general shear failure for some cases and
failure due to excessive settlement for other cases. Numerical
analyses were carried out to determine the possible failure
mode; however, because of great varieties in the properties of
rock masses, using numerical analyses may not be possible for
practical applications. Therefore, it would be better to derive
simple charts or tables in this regard. For achieving this purpose,
an extensive field test seems to be necessary.
References
[1] Serrano, A. and Olalla, C. ‘‘Ultimate bearing capacity of an anisotropic
discontinuous rock mass, part I: basic modes of failure’’, Int. J. Rock Mech.
Min. Sci., 35(3), pp. 301–324 (1998).
[2] Serrano, A. and Olalla, C. ‘‘Ultimate bearing capacity of an anisotropic
discontinuous rock mass, part II: determination procedure’’, Int. J. Rock
Mech. Min. Sci., 35(3), pp. 325–348 (1998).
[3] Sutcliffe, D.J., Yu, H.S. and Sloan, S.W. ‘‘Lower bound solutions for bearing
capacity of jointed rock’’, Comput. Geotech., 31, pp. 23–36 (2004).
[4] Yang, X.L. and Yin, J.H. ‘‘Upper bound solution for ultimate bearing capacity
with a modified Hoek–Brown failure criterion’’, Int. J. Rock Mech. Min. Sci.,
42, pp. 550–560 (2005).
 M. Imani et al. / Scientia Iranica, Transactions A: Civil Engineering 19 (2012) 1411–1421
[5] Merifield, R.S., Lyamin, A.V. and Sloan, S.W. ‘‘Limit analysis solutions for
the bearing capacity of rock masses using the generalised Hoek–Brown
criterion’’, Int. J. Rock Mech. Min. Sci., 43, pp. 920–937 (2006).
[6] Saada, Z., Maghous, S. and Garnier, D. ‘‘Bearing capacity of shallow
foundations on rocks obeying a modified Hoek–Brown failure criterion’’,
Comput. Geotech., 35, pp. 144–154 (2008).
[7] Maghous, S., de Buhan, P. and Bekaert, A. ‘‘Failure design of jointed rock
structures by means of a homogenization approach’’, Mech. Cohesive Frict.
Mater., 3, pp. 207–228 (1998).
[8] Halakatevakis, N. and Sofianos, A.I. ‘‘Strength of a blocky rock mass based
on an extended plane of weakness theory’’, Int. J. Rock Mech. Min. Sci., 47,
pp. 568–582 (2010).
[9] Serrano, A. and Olalla, C. ‘‘Allowable bearing capacity of rock foundations
using a non-linear failure criterium’’, Int. J. Rock Mech. Min. Sci. Geomech.
Abstr., 33, pp. 327–345 (1996).
[10] Frank, R., Bauduin, C., Driscoll, R., Kavvadas, M., Ovesen, N.K., Orr, T. and
Schuppener, B., Designer’s Guide to EN 1997-1, Eurocode 7: Geotechnical
Design-General Rules, first ed., Thomas Telford, London (2004).
[11] Davis, E.H. ‘‘Theories of plasticity and the failure of the soil masses’’, In Soil
Mechanics: Selected Topics, I.K. Lee, Ed., Butterworths, London (1968).
[12] Maghous, S., Bernaud, D., Freard, J. and Garnier, D. ‘‘Elastoplastic behavior
of jointed rock masses as homogenized media and finite element analysis’’,
Int. J. Rock Mech. Min. Sci., 45, pp. 1273–1286 (2008).
[13] Imani, M., Sharifzadeh, M., Fahimifar, A. and Haghparast, P. ‘‘A characteris-
tic criterion to distinguish continuity of rock masses applicable to founda-
tions’’, Proc. 45th US Rock Mech/ Geomech. Symposium, San Francisco, USA,
ARMA-11-508 (2011).
[14] Alehossein, H., Carter, J.P. and Booker, J.R. ‘‘Finite element analysis of rigid
footings on jointed rock’’, Proc. 3rd Int. Conf. Computational Plasticit, 2, pp.
935–945 (1992).
Meysam Imani obtained his B.S. in civil engineering from Tehran University in
2004. He received his M.Sc. and Ph.D. degrees in soil mechanics and foundation
engineering from Amirkabir University of Technology (Tehran Polytechnic)
in 2006 and 2012, respectively. He published several papers in the field of
rock foundations in different journals and international conferences. He has
1421
delivered some lectures on Soil Mechanics and Foundation Engineering in
Amirkabir University of Technology, Garmsar Campus and also in Islamic
Azad University. The main topics of his research interests are: Analytical
Methods in Geomechanics, Upper and Lower Bound Methods of Limit Analysis,
Rock Foundations, Finite Element Modeling of Soil–Structure Interaction and
Mechanized Tunneling. He has also made contributions as a geotechnical
consultant in practical projects such as Mechanized Tunneling, Concrete Dams
and Cement Factories.
Ahmad Fahimifar studied civil engineering at Iran University of Science and
Technology, gaining B.S. degree in 1976. After a period in industry, he continued
his M.S. in the University of Birmingham (United Kingdom). He was awarded a
Ph.D. degree by the University of Newcastle Upon Tyne (United Kingdom) in
1990 for his work on behavior of jointed rocks.
He began his academic activities in the Department of Civil Engineering
at Amirkabir University of Technology in 1990, and appointed as the Head
of Department in 1992. He has delivered many lectures on Rock Mechanics
and Tunnel Engineering. In 2006 he was appointed as Professor at Amirkabir
University. He is a member of International Society for Rock Mechanics, the
founder and a member of Iranian Society for Rock Mechanics, the founder and a
member of Iran Tunneling Association, and the founder and a member of Iranian
Society of Civil Engineers. He is a member of some Iranian research journals
including: Amirkabir Journal of Science and Technology, ASAS Journal of Civil
Engineering, and Journal of Engineering Geology and the Environment.
Mostafa Sharifzadeh has a B.S. degree in Mining Engineering (1993-Kerman
University), an M.S. degree in Engineering Rock Mechanics (1997-Amirkabir
University of Technology, Tehran, Iran), and a Ph.D. degree in Environmental
Geotechnology (2005-Kyushu University, Japan). Currently he is Associate
Professor in Amirkabir University of Technology (Tehran Polytechnic). He is
the author of papers on various subjects in rock mechanics, rock engineering,
geotechnical engineering, tunnel engineering and numerical modeling. He
has published over 200 publications including 3 books, 35 national and
international indexed journal paper, and 100 conference papers in national
and international conferences, 25 reports, 45 guideline and standards. He has
successfully supervised over 20 undergraduate and 45 graduate students’ thesis
in master and doctoral level.